國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
multiple senses of of be linked in a network are therefore the topics under investigation for the current study.
1.2 TheOf -construction and Semantic Relations
This study investigates a complex noun phrase (CNP) containing of, designated as an of -construction, i.e., constructions in the form of [N1 of N2], where N1 and N2 are noun phrases (NPs). We focus on English academic writing. Corpus examples taken from the academic subcomponent of BNC are shown in (1.1) with the of -construction underlined.
(1.1) a. James sought to reassert the divine right of kings, and Parliament combined against him. (J57_675) possessive
b. Some idea of the effect can be gained by depressing the surface of a table tennis ball with the thumb. (J0T_360) part-whole
c. Bile acids are necessary for the absorption of dietary fats and cholesterol from the intestine. (HU2_7495) nominalization
These examples demonstrate three distinct functions expressed by of -constructions.
First of all, the of -construction in (1.1a) exemplifies a possessive relationship between the divine right and kings which, in general, has been considered as a prototypical function of the s-genitive construction which, under certain circumstances, can alternate with the of -construction (e.g., Kreyer, 2003; Quirk, Greenbaum, Leech &
Svartvik, 1985; Stefanowitsch, 2003). The second example (1.1b) epitomizes a cognitively fundamental relationship, the part-whole relation (e.g., Chaffin &
Herrmann, 1987), signaled by the NPs the surface and a table tennis ball. The third example in (1.1c) contains the nominalization of a phenomenon, which is also commonly associated with the of -construction, particularly prevalent in academic discourse (e.g., Halliday & Martin, 1993). The variety of of -constructions illustrated here demonstrates the complexity of the constructions due to the wide range of meanings involved. Furthermore, the relationship between the two NPs, which is
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
referred to as a semantic relation throughout this study, appear to facilitate our comprehension of the of -construction.
Semantic relations, according to Chaffin and Herrmann (1987), is rich in natural language in terms of their varieties. While a semantic relation can be described between nouns (e.g., a hyponymic relation between animal and dog), it can also describe the relations between adjectives (e.g., an antonymic relation between large and small) or between an adjective and a noun (e.g., a property relation between heavy and weight). In terms of the of -construction, the semantic relation is between two nouns, N1 and N2. Fellbaum (2002) defines the term semantic relations in her work on troponomy. She describes semantic relations as to be derived from the situation where
“If one examines the lexicalized concepts in relation to one another, it becomes clear that they differ in systematic ways that are characterizable in terms of similarities or contrasts” (p.23). For the purpose of this study, we adopt Fellbaum’s use of lexicalized concept to denote semantic relations as the association between meanings of words in context. Although the of -construction is well considered mainly from the literature on genitive alternation between of and s-genitive (see Rosenbach, 2014 for a comprehensive review), there have been comparatively fewer cases that examine the semantic relations of the of -construction in detail (e.g., Hasselgård, 2016; Langacker, 1982, 1999; Lindstromberg, 2010; Schonthal, 2016). For example, Lindstromberg (2010) uses the term integration to classify various types of of -construction including part-whole (e.g., the eye of a storm), product-source (e.g., a result of hard work), action-agent (e.g., the howling of dogs), action-patient (e.g., the delivery of goods), and example-type (e.g., a sample of tissue). This approach, which considers the relationship between the two NPs, is very similar to the research in information technology where automatic processing of human language is the goal. The work of Moldovan, Badulescu, Tatu, Antohe, and Girju (2004) represents one of them. In their study, automatic labelling of semantic relations on a variety of NP-associated phrases such as noun-noun compounds, adjective phrases, s-genitives, and of -constructions are considered. Out of a total number of 35 relations, they suggest that there are 21
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
relations found in the of -construction. While the types of semantic relations have been established, what remains to be done is to determine how different types of semantic relations are used by language users and with what purposes.
Furthermore, although of ranks the top most frequent preposition in a variety of text types (e.g., Biber & Gray, 2010), the of -construction has rarely been considered in its own right in English for Academic Purposes (EAP) where instructions for L2 novice writers in addition to language research are both covered (Hyland & Shaw, 2016). This may be due to the misconception that of is a meaningless function word on the one hand, and on the other hand that it is well-established that of expresses a fundamental concept of part-whole relation and it is commonly viewed as the alternative of the genitive, and nothing else. While the argument that most function words are inherently low in semantic content has no longer been held to be true in the cognitive linguistic paradigm (Langacker, 1982, 1999), like many other preposition words, of is gaining its status in meaning content. Other fields in linguistics, however, have investigated of in more detail. For example, Renouf and Sinclair (1991) demonstrate that highly frequent grammatical words such as to and of form collocational frameworks that associate with particular groups of words. They examine several word sequences containing to and of that can be filled with various lexical items in between two fixed words such as [be+?+to], [too+?+to], [a+?+of ], [an+?+of ], [for+?+of ], [had+?+of ], and [many+?+of ] and found that the slot in [a+?+of ] consists of a large number of quantity nouns such as couple, series, and lot, whereas [an+?+of ] tends to be occupied by a number of nominalized words such as extension and indication.1 This study and Sinclair’s (1991) work on the semantic head of of -constructions are devoted to demonstrate the complexity that of is woven in the semantic network of collocation. Another data-driven work that also demonstrates a rich system of of -constructions is perhaps Francis, Hunston, and Manning’s (1998) contribution to the pattern [N of N]. A total of 39 subcategories have been identified based on the semantics of the first N including subcategories such as the
1Each question mark occupies one slot for a word.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
announcement group (e.g., charges of murder, accusations of being ‘soft’), the sign group (e.g., a neighbourly gesture of goodwill, the proof of a good song), and the cause and result group (e.g., an agent of change, the secret of your success). This work illustrates a multitude of meanings of of, while Renouf and Sinclair demonstrate a multitude of phrasal structures involving the word of.
Polysemous words like English prepositions are difficult to attain for second/foreign language (L2) learners of English (e.g., Cho, 2010). In fact, prepositional studies based on error analysis, such as those using an error-tagged learner corpus, have demonstrated L2 learners’ struggle with English prepositions including of (e.g., Chan, 2010; Chuang & Nasi, 2006). Although the productivity of the of -construction has not drawn much attention on how it can be taught effectively in teaching, the fact that it is characteristic of academic writing has been confirmed (e.g., Biber, et al., 1999; Halliday & Martin, 1993; Halliday, 2004, Ventola, 1996), particularly in conjunction with nominalization (such as deverbal nouns, e.g., pronunciation, or deadjectival nouns, e.g., effectiveness) or postmodifier complexes (e.g., in a number of ways) (Biber, et al., 1999: 642). In an investigation into lexical density in English academic writing, Ventola (1996) notices the increased structural complexity of nominal groups by using multiple of -constructions with the center nominal group to be surrounded by both a premodifier and a postmodifier.
(1.2) the solution of the stationary distribution of the stochastic cash-management model (taken from Ventola, 1996: 180)
In Example (1.2), there are three nominal groups (underlined) headed by solution, distribution, and model. Ventola suggests that the interpretation for this complex of -construction can be followed in two steps. First, the very last nominal phrase, the stochastic cash-management model, qualifies the second nominal phrase, the stationary distribution. In turn, these two phrases function together to serve as a qualifier for the first nominal group, the solution. The author states that the last of -construction “actually functions as a Qualifier of a Qualifier” (p.180) and packaging of information of this kind is particularly common in academic writing.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
However, considering an of -phrase (i.e., of + NP) as a qualifier of the previous NP might face problems when we have phrases like the disruptive power of certain kinds of language (5491-J2K). It would be awkward to say that language qualities certain kinds. Therefore, a different perspective perhaps is necessary to re-examine this type of constructions.
Another important source of research that demonstrates the close association of of -constructions in academic writing is that by Biber et al. (1999). In a large-scale, corpus-based investigation, Biber et al. examine of by comparing its use in academic prose with another register, conversation. The authors adopt the analysis of lexical bundles, defined as “the sequences of words that most commonly co-occur in a register” or “extended collocations” (p.989), to investigate the patterns of word strings.
The result not only reveals areas of contrast between the two registers, it also shows that a fair number of these lexical bundles in academic prose contain an of -phrase fragment (e.g. the end of the, the position of the, as a result of, and about the nature of ). These lexical bundles appear to be part of an NP or a prepositional phrase which are common in academic writing, whereas lexical bundles in conversation are more commonly found in verb phrases. This approach exposes a list of collocating word strings containing of -constructions by taking into account of register variations.
In general, previous studies have demonstrated that the of -construction is not only complicated in terms of its multiplicity in meanings but also its presence in various syntactic positions. Just like any nominal group, the of -construction can appear in the subject, object, and complement not only as a noun phrase but also as a prepositional phrase. The results of Biber et al.’s study show that the of -construction seems to have an affinity with prepositional phrases. The purpose of this study is therefore to identify how these of -containing bundles work in various syntactic positions and to what extent they are used as a prepositional phrase.
While the discussion above highlights some fruitful findings, teaching intervention seems to remain difficult to be constructed. One of the hurdles is perhaps the lack of sufficient information on the second nominal in relation to the first. For example,
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Biber et al.’s work informs us about the likely collocates of N1 (e.g., the end of, as a result of ), but an expectation of an automatic filling in appropriate words at N2 might cause a problem for some L2 learners as they might not have sufficient amount of target input yet. In other words, a comprehensive analysis that considers the entire set of of -constructions could perhaps benefit L2 learners by raising their awareness of of ’s complexity.